Classification society standard requires redundancy in floating vessels with thruster assisted mooring system, but it is not avoidable for floating vessels to drive off in an instant of a mooring line failure such as line breakage or dragging of anchor. Such a failure might cause serious hazard by consequent mooring line breakages. Agile response of assisted thrusters in case of mooring line failure is essential to minimize excursion and prevent consequent hazard. Conventional controller based on the moored vessel's intact dynamics is not proper to handle the configuration and dynamic characteristic changed by a failure and it might cause time delay to be stable and recover the vessel's position. This paper presents a neural network based adaptive control method that guarantees stable control performance even with mooring line failure. Neural network based adaptive control can achieve control objectives in the presence of modeling uncertainty. The advantage of the proposed control method is that the controller does not require dynamical model of failure condition. In the case of mooring line failure, the neural network regards configuration change as added modeling uncertainty. Numerical simulations of a semi-submersible type vessel were performed for the validation. Proposed method has robust control performance in the presence of a failure. The results show that the neural network based adaptive control method enables a vessel to response timely and recover the position effectively.